
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns 
import pylab
from scipy import stats
from scipy.stats import norm, probplot
from pandas_profiling import ProfileReport
import statsmodels.api as sm

from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import plot_confusion_matrix, plot_roc_curve
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier

# Importing the SMOTE-NC
from imblearn.over_sampling import SMOTENC


# Dataset Import and working copy generation
raw_data = pd.read_csv('./healthcare-dataset-stroke-data.csv')
df = raw_data.copy()
df.head(10)


profile = ProfileReport(df, title="Pandas Profiling Report", explorative=True)
profile


# Remove bmi's na
df = df.dropna()

# Remove id column
df = df.drop(columns=['id'])

# Remove 'Other' sex register 
df = df[df.gender != 'Other']

# Last show a summary of clean data
profile = ProfileReport(df, title="Pandas Profiling Report", explorative=True)
profile


fatties = df[df.bmi > 47.5]
fatties.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 110 entries, 21 to 5057
Data columns (total 11 columns):
 #   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  
 0   gender             110 non-null    object 
 1   age                110 non-null    float64
 2   hypertension       110 non-null    int64  
 3   heart_disease      110 non-null    int64  
 4   ever_married       110 non-null    object 
 5   work_type          110 non-null    object 
 6   Residence_type     110 non-null    object 
 7   avg_glucose_level  110 non-null    float64
 8   bmi                110 non-null    float64
 9   smoking_status     110 non-null    object 
 10  stroke             110 non-null    int64  
dtypes: float64(3), int64(3), object(5)
memory usage: 10.3+ KB


stroked_fatties = fatties[fatties.stroke==1]
stroked_fatties.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 2 entries, 21 to 113
Data columns (total 11 columns):
 #   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  
 0   gender             2 non-null      object 
 1   age                2 non-null      float64
 2   hypertension       2 non-null      int64  
 3   heart_disease      2 non-null      int64  
 4   ever_married       2 non-null      object 
 5   work_type          2 non-null      object 
 6   Residence_type     2 non-null      object 
 7   avg_glucose_level  2 non-null      float64
 8   bmi                2 non-null      float64
 9   smoking_status     2 non-null      object 
 10  stroke             2 non-null      int64  
dtypes: float64(3), int64(3), object(5)
memory usage: 192.0+ bytes


# Bar plots as function of stroke status
categs = ['gender',
          'hypertension',
          'heart_disease',
          'ever_married',
          'work_type',
          'Residence_type',
          'smoking_status']

sns.set_theme(style='darkgrid', palette='pastel')

for cat in categs:
    ax = sns.countplot(data=df, x=cat, hue='stroke')
    plt.show()


# Histograms as function of stroke status
nums = ['age',
        'avg_glucose_level',
        'bmi']

for num in nums:
    ax = sns.histplot(data=df, 
                      x=num,
                      multiple='stack',
                      bins=20,
                      stat='density',
                      hue='stroke')
    plt.show()


measurements = df['age']
probplot(measurements, dist="norm", plot=pylab)
pylab.show()


measurements = df['bmi']
probplot(measurements, dist="norm", plot=pylab)
pylab.show()


measurements = df['avg_glucose_level']
probplot(measurements, dist="norm", plot=pylab)
pylab.show()


# Splitting attributes and labels
X = df.drop(['stroke'], axis=1)
y = df['stroke']

# One hot encoding of categorical attributes
X = pd.get_dummies(X, prefix=['gndr', 'marr', 'wrk', 'res', 'smk'])


# Selection of numeric attributes
X_num = X[['avg_glucose_level', 'bmi']]

# Standardization of numeric variables
scaler = StandardScaler()
X_num = scaler.fit_transform(X_num)

# Dropping of old numerics
X = X.drop(['avg_glucose_level', 'bmi'], axis=1)

# Conversion to np arrays and concatenation
X_np = X.to_numpy()
X_np= np.concatenate((X_np, X_num), axis=1)
 
# Ceating train and test sets
X_train, X_test, y_train, y_test = train_test_split(X_np, y, test_size=0.3, stratify=y, random_state=2021)


# Create the oversampler 
smotenc = SMOTENC([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17], random_state=1963)

# We synthesize new ,  values
X_smt, y_smt = smotenc.fit_resample(X_train, y_train)

# We check average statistics of labels
print('\n\n',y_smt.describe())


 count    6578.000000
mean        0.500000
std         0.500038
min         0.000000
25%         0.000000
50%         0.500000
75%         1.000000
max         1.000000
Name: stroke, dtype: float64


clf = LogisticRegression(max_iter=1000)
clf.fit(X_train, y_train)
plot_confusion_matrix(clf, X_test, y_test, cmap='OrRd')
plot_roc_curve(clf, X_test, y_test)

<sklearn.metrics._plot.roc_curve.RocCurveDisplay at 0x7f72a759d520>


# Logistic classification with enlarged dataset
clf_smt = LogisticRegression(max_iter=1000)
clf_smt.fit(X_smt, y_smt)
plot_confusion_matrix(clf_smt, X_test, y_test, cmap='OrRd')
plot_roc_curve(clf_smt, X_test, y_test)

<sklearn.metrics._plot.roc_curve.RocCurveDisplay at 0x7f72a7416790>


# Function definition for plotting results
def viz_results(model):
    model.fit(X_smt, y_smt)
    plot_confusion_matrix(model, X_test, y_test, cmap='OrRd')
    plot_roc_curve(model, X_test, y_test)


# Models selection
KNN = KNeighborsClassifier()
VecMa = SVC(gamma='auto')
DecTree = DecisionTreeClassifier(max_depth=5)
models = [KNN, VecMa, DecTree]
for model in models:
    viz_results(model)
#viz_results(DecTree)

	id	gender	age	hypertension	heart_disease	ever_married	work_type	Residence_type	avg_glucose_level	bmi	smoking_status	stroke
0	9046	Male	67.0	0	1	Yes	Private	Urban	228.69	36.6	formerly smoked	1
1	51676	Female	61.0	0	0	Yes	Self-employed	Rural	202.21	NaN	never smoked	1
2	31112	Male	80.0	0	1	Yes	Private	Rural	105.92	32.5	never smoked	1
3	60182	Female	49.0	0	0	Yes	Private	Urban	171.23	34.4	smokes	1
4	1665	Female	79.0	1	0	Yes	Self-employed	Rural	174.12	24.0	never smoked	1
5	56669	Male	81.0	0	0	Yes	Private	Urban	186.21	29.0	formerly smoked	1
6	53882	Male	74.0	1	1	Yes	Private	Rural	70.09	27.4	never smoked	1
7	10434	Female	69.0	0	0	No	Private	Urban	94.39	22.8	never smoked	1
8	27419	Female	59.0	0	0	Yes	Private	Rural	76.15	NaN	Unknown	1
9	60491	Female	78.0	0	0	Yes	Private	Urban	58.57	24.2	Unknown	1

Contribucions	Signa
Investigació prèvia	jgualob, dhu
Redacció de les respostes	jgualob, dhu
Desenvolupament codi	jgualob, dhu

Predicció accidents vasculars cerebrals¶

NOM I COGNOMS: Jordi Gual Obradors | Daniel Lijia Hu¶

1. Descripció del dataset. Perquè és important i quina pregunta/problema pretén respondre?¶

Context:¶

Informació dels atributs.¶

2. Integració i selecció de les dades d’interès a analitzar¶

3. Neteja de les dades.¶

3.1. Les dades contenen zeros o elements buits? Com gestionaries aquests casos?¶

3.2. Identificació i tractament de valors extrems.¶

4. Anàlisi de les dades.¶

4.1. Selecció dels grups de dades que es volen analitzar/comparar (planificació dels anàlisis a aplicar).¶

4.2. Comprovació de la normalitat i homogeneïtat de la variància.¶

6. Resolució del problema. A partir dels resultats obtinguts, quines són les conclusions? Els resultats permeten respondre al problema?¶